Does anyone have any FFT plots from real Class-AB tube amps?

[bigwill]

I have a nice little amp going in SPICE - it simulates pretty well. Below 10W the distortion is very low both in amplitude and order. Above about 10W, the amp is no longer in Class-A. The output sine looks very clean but the FFT plot reveals a horrible spray of harmonics, starting at 50-60dB down and falling off with frequency.

Is the real life picture this bad for AB transitions? It seems like NFB would have a hard job clearing up the mess I see in simulation. I've always heard the AB transition wasn't so harsh with valves..

Can anyone offer an insight as to what a real tube amp looks like while running strongly in Class-AB in terms of spectra?

[teemuk]

And that would be a real tube amp with a THD of...?


SPICE simulations may introduce harmonics due to e.g. waveform compression (that compresses simulation data enhancing speed) or because the time frame to plot the wave and perform a FFT analysis isn't set up properly.

[SGregory]

This is the spectrum of my OPUS at 1 Watt and 20 Watt. Not sure if that is what you mean by a "REAL" amp or not. But it physically exists and these are the measurements I took of it. I also am interested in some "REAL Commercial" amp spectrums as well so maybe somebody would post those spectrums.

It looks ugly but it is the best sounding amp I have ever built.

Is your sims using feedback?

[bigwill]

Originally Posted by SGregory

Not sure if that is what you mean by a "REAL" amp or not.

Physically existing

Originally Posted by SGregory

Is your sims using feedback?

A lot of local feedback and a little global feedback. This seems to work very well during Class-A, in fact the results look very much like yours, low order/magnitude at low wattage and harmonic "grass" in Class-AB, although I must say your spectrum looks excellent, with everything above the 5th harmonic below 100dB!

[john curl]

SGregory, your plots look normal to me. I bet that it sounds pretty good, also.

[SGregory]

Okay, here is one at 40Watt. Not quite as pretty.

Have you looked at B+ regulation for the driver circuits? I am getting ready to try it out once the parts arrive.

[SGregory]

Thanks John, the sound is wonderful. I have rolled a few tubes on the other channel and was able to get the even harmonics to be more dominant at a slight thd penalty. I'll post that on the OPUS thread in a little while.

I, like bigwill, am curious in reducing the "grass" at higher powers further. In my case with minimum feedback.

[Globulator]

I wonder if the higher harmonics is due to the global NFB, as the power increases and the corrections get bigger, all those just-not-in-time corrections cycle around and around spread all over.

I'd be interested to see what you get without any NFB (maybe just local FB).

How do you measure the FFT by the way.

[bigwill]

Here's the FFT from my SIMULATED output stage / driver stage combined running at 10V pk-pk into 8 ohms (a little over a watt)

[bigwill]

And here's the plot of around 13W into 8ohms (no global feedback in both images), only just a touch into AB I'm hoping real life isn't as bad as this: